情感计算技术如何推动犯罪风险评估工具的发展？

摘 要 随着近年来人工智能深度学习技术的发展,情感计算与人格计算技术日渐成熟,在许多实际应用场景中取得了良好的表现,当前人工智能情感计算技术应用于犯罪风险评估领域,能够有效解决目前主流的风险评估工具难以解决的个体内差异性的预测因子评估问题以及被测评参与者因社会赞许性而导致结果失真的问题。本文在详细阐述目前主流评估工具的局限性基础上,详细阐述了以情感计算技术为支撑的动态风险评估工具的设计思路、目前已有的技术方案以及设计细节的理论依据,在此基础上最后提出以人工智能技术为支撑的新型评估工具的未来发展方向。     

Factors associated with experts' opinions regarding criminal responsibility in The Netherlands

In many jurisdictions, offenders need to have freely chosen to commit their crimes in order to be punishable. A mental defect or disorder may be a reason for diminished or total absence of criminal responsibility and may remove culpability. This study aims to provide an empirically based understanding of the factors on which experts base their judgements concerning criminal responsibility. Clinical, demographic and crime related variables, as well as MMPI-2 profiles, were collected from final reports concerning defendants of serious crime submitted to the observation clinic of the Dutch Ministry of Justice for a criminal responsibility assessment. Criminal responsibility was expressed along a five-point scale corresponding to the Dutch legal practice. Results showed that several variables contributed independently to experts' opinions regarding criminal responsibility: diagnosis (Axis I and II), cultural background, type of weapon used in committing the crime, and whether the defendant committed the crime alone or with others. In contract to jurisdictions involving a sane/insane dichotomy, the Dutch five-point scale of criminal responsibility revealed that Axis II personality disorders turned out to be mostly associated with a diminished responsibility. MMPI-2 scores also turned out to have a small contribution to experts' opinions on criminal responsibility, independently of mere diagnostic variables. These results suggest that experts base their judgements not only on the presence or absence of mental disorders, but also on cultural and crime related characteristics, as well as dimensional information about the defendant's personality and symptomatology.     

Further examination of video‐based preference assessments without contingent access

The first purpose of this study was to attempt to replicate previous findings that video‐based preference assessments without access to selected stimuli may accurately predict relative reinforcing efficacy of stimuli. To do this, we conducted a concurrent operant reinforcer assessment in which we evaluated the relative reinforcing value of highly preferred and less preferred items identified in a video‐based preference assessment. The second purpose of this study was to begin to evaluate the potential behavioral mechanisms responsible for the validity of this assessment. To conduct this analysis, we evaluated the relative reinforcing value of those same stimuli depicted in video format and then compared results to results obtained during the reinforcer assessment for tangible stimuli. For all five participants, stimuli identified as highly preferred functioned as reinforcers, and four of five participants, responding during the reinforcer assessment was similar in the presence of tangible stimuli and videos depicting those stimuli.     

Ecological assessments of community disorder: Their relationship to fear of crime and theoretical implications

Researchers suggest that fear of crime arises from community disorder, cues in the social and physical environment that are distinct from crime itself. Three ecological methods of measuring community disorder are presented: resident perceptions reported in surveys and on-site observations by trained raters, both aggregated to the street block level, and content analysis of crime- and disorder-related newspaper articles aggregated to the neighborhood level. Each method demonstrated adequate reliability and roughly equal ability to predict subsequent fear of crime among 412 residents of 50 blocks in 50 neighborhoods in Baltimore, MD. Pearson and partial correlations (controlling for sex, race, age, and victimization) were calculated at multiple levels of analysis: individual, individual deviation from block, and community (block/neighborhood). Hierarchical linear models provided comparable results under more stringent conditions. Results linking different measure of disorder with fear, and individual and aggregated demographics with fear inform theories about fear of crime and extend research on the impact of community social and physical disorder. Implications for ecological assessment of community social and physical environments are discussed.     

THE FUNCTIONS OF ASSESSMENT: IMPLICATIONS FOR SELECTION AND DEVELOPMENT OF DEVICES FOR ASSESSING REPERTOIRES IN CLINICAL,EDUCATIONAL, AND OTHER SETTINGS

An attempt is made to identify the many different functions that assessment of an individual's repertoire can serve. Implications of these functions for the character of and evidence about assessment devices are suggested. The functions fall into two general groups, those which influence decisions regarding an individual learner, and those which influence policy, program development, and scientific knowledge. The first group of functions is presented in a rough chronological sequence such that they form a “behavioral assessment funnel,” beginning with functions involving broad-band assessment to identify likely persons and skill areas, and narrowing to the precise pinpointing, monitoring, and follow-up functions. The contribution of behavior analysis and behavior therapy to assessment methodology in this sequence is identified as well as the areas where more traditionally conceived methods are still useful. The second group of functions and behavioral contributions to it are then discussed.     

履历数据测评的效度分析

履历数据是人员测评的重要手段, 但其组织情景性限制使得国外同类工具不能适用于国内人事选拔, 国内也缺乏实证研究验证其信度、效度。本研究结合某通信企业的招募选拔工作, 开发履历数据分析工具, 选取250名应聘者的履历数据、一般认知能力、大五个性和面试结果数据, 通过效标关联效度、增量效度思想, 分析履历数据有效性。结果表明, 以面试结果为效标, 履历数据具有良好的效标关联效度, 与其他测评工具组合使用时有良好的增量效度。     

TWO MEASURES OF PREFERENCE DURING FORCED-CHOICE ASSESSMENTS

A forced-choice preference assessment was conducted in which two dependent measures were used to select preferred stimuli: (a) approach responding and (b) latency to the first aberrant response. Stimuli identified as preferred based on both dependent measures were then evaluated during treatment. The results suggested that latency may be a useful measure in the selection of preferred stimuli during forced-choice assessments.     

the relationship between elements and manifestations of low self-control in a general theory of crime: two comments and a test

I discuss two issues about the empirical assessment of Gottfredson and Hirschi's (1990) general theory of crime. First, regarding the internal validity of corresponding causal models, I argue that more attention must be paid to the distinction between independent and dependent variables (elements versus manifestations of low self-control). Second, regarding the interpretation of statistical results, I argue that researchers should not lose sight of the distinction between statistical and substantive significance. I also report a test of the theory based on data from a representative sample of high school seniors in the United States. The results of the test show that low self-control is a correct and certain, yet weak, predictor of certain deviant behaviors (skipping class, traffic violations, cigarette smoking, and the use of alcohol and marijuana).     

Immigrant-Friendly and Unfriendly Cities: Impacts on the Presence of a Foreign-Born Population and City Crime

Immigration continues to be an issue in the USA. In the absence of substantive Federal law, some local governments have passed ordinances related to immigrants living in their community. Some of these have had pro-immigrant orientations and some have had anti-immigrant orientations. In the literature, these types of policies have been found to have mixed effects on immigrant decisions to live in those communities. These cities have passed these ordinances in order to attract or repel them because of perceived impacts on city crime among other reasons. This study is the first assessment on whether these policies have attracted or repelled the foreign-born population in these cities, and it contributes to the ongoing literature on whether that foreign-born population is having an effect on city crime. It was found that these policies are not having a significant effect on attracting or repelling immigrants while the impact of this population on city crime is significantly negative. Further research is recommended on this important topic.     

Involving Police Departments in Early Awareness of Concussion Symptoms during Domestic Violence Calls

ABSTRACT

Police officers around the world respond to and investigate calls regarding domestic violence (DV) daily. Police departments operate with standard protocols, particularly when engaging in investigations that involve allegations of strangulation or sexual assault. Operating under advisement of the Maricopa County Attorney’s Office, the Mesa Police Department (Mesa, AZ) has established protocols for detecting, recording, and prosecuting DV cases that involve strangulation and sexual assault. Allegations of strangulation (defined as impeded breathing) or sexual assault prompt officers to offer a forensic nursing exam (FNE) combined with strangulation treatment by forensic nurses at the Mesa Family Advocacy Center. Recognizing the potential for head injury to the assault victim in all DV situations, including intimate partner violence (IPV), the police department has added concussion-awareness training, as well as a point-of-incident investigative tool for its officers to record neurological function of the victim. Officers were instructed to use the ConQVerge device to measure and record the Near Point of Convergence (NPC) as a test of neurological impairment in suspected head injuries. In this article, we discuss the challenges and opportunities for assisting victims of DV strangulation and sexual assault (including non-DV sexual assault) with on-site assessment and consent for further medical assessment and treatment. Additionally, rates of domestic assault victims that report a crime, but decline to follow through with forensic medical tests, are reported for the first time. Lessons learned from the project that inform strategic operations in this space are offered to other agencies prior to the implementation of similar procedures.     

ON THE RELATION BETWEEN REINFORCER EFFICACY AND PREFERENCE

Most research on stimulus preference and reinforcer assessment involves a preference assessment that is followed by a reinforcer assessment. Typically, the most and least preferred stimuli are tested as reinforcers. In the current study, we first quantified the reinforcing efficacies of six food items and then assessed relative preference for each item. Relative preference ranking and reinforcer efficacies showed almost perfect concordance for 1 participant and partial concordance for the other. Discordance tended to occur with the weakest reinforcers.     

Assessing Violence Risk Among Female IPV Perpetrators: An Examination of the B-SAFER

Intimate partner violence (IPV) is a global problem and one in which frontline assessment and management falls primarily to police. Although IPV is often conceptualized as a male-perpetrated crime, evidence substantiates female IPV perpetration and increased arrest rates, raising important issues for police. This article examines police use of the Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER; Kropp, Hart, & Belfrage, 2005, 2010), a violence risk assessment tool for IPV. The B-SAFER was used to assess and manage 52 women arrested for IPV. When compared to Belfrage and Strand (2008), who examined men arrested for IPV in the same sample, women possessed fewer risk factors. Risk factors were related to summary risk judgments, although differences existed between genders and risk judgments were not related to management recommendations. Results suggest that risk factors, in addition to those in the B-SAFER, are required to assess risk for female IPV.     

Ambient Temperature and Violent Crime1

Baron and Ransberger (1978) argue that civil violence increases as temperature rises into the mid 80s, and then decreases as temperatures rise further. Two experiments test this hypothesis using data on temperature and the incidence of crime for summer months in two midwestern cities. The crime data were divided into violent and nonviolent crimes, and then correlated with the maximum, minimum, and average temperatures, and several humidity measures. Violent crime correlated significantly with temperature; nonviolent crime did not. This relationship was linear in the first study, but curvilinear in the second. Inspection of the data suggests that the incidence of aggressive behavior (i.e., violent crime) does not drop off in the mid-80s as laboratory findings and Baron and Ransberger's results would suggest, but continues to increase as temperatures rise into the 90s. The mean incidence of violent crime was higher for days in the 90s than for days in the 80s. A hypothesis for resolving this contradiction between real world and laboratory findings is discussed.     

Teacher report and direct assessment of preferences for identifying reinforcers for young children

Teachers were asked to identify and rank 10 preferred stimuli for 9 toddlers, and a hierarchy of preference for these items was determined via a direct preference assessment. The reinforcing efficacy of the most highly preferred items identified by each method was evaluated concurrently in a reinforcer assessment. The reinforcer assessment showed that all stimuli identified as highly preferred via the direct preference assessment and teacher rankings functioned as reinforcers. The highest ranked stimuli in the direct assessment were more reinforcing than the teachers' top-ranked stimuli for 5 of 9 participants, whereas the teachers' top-ranked stimulus was more reinforcing than the highest ranked stimulus of the direct assessment for only 1 child. A strong positive correlation between rankings generated through the two assessments was identified for only 1 of the 9 participants. Despite poor correspondence between rankings generated through the teacher interview and direct preference assessment, results of the reinforcer assessment suggest that both methods are effective in identifying reinforcers for toddlers.     

russian organized crime: the continuation of an american tradition

This research examines the emergence of Russian émigré crime in the United States to determine whether such crime is representative of organized crime in the classic understanding of that term. The prior development of organized crime theory based upon the criminal progression of Italian-American organized crime families guides the analysis of criminal activity by the Russian émigrés in order to maintain some sense of boundary maintenance between events and issues pertinent to the phenomenon of organized crime. With a model for Russian émigrécrime in the United States established, the research examines the underlying characteristics of the emergent Russian criminality with a goal of determining which theoretical model, if any, this social system of criminality fits best. Case studies of recent motor fuel tax frauds perpetrated in the metropolitan New York City and Los Angeles areas serve as examples from which Russian émigré crime is examined, studied, and compared. An assessment is made as to whether the criminality of the émigré Russians can be correctly identified as being representative of organized crime in the traditional understanding of this phenomenon. The analysis demonstrates that the criminality of the Russian e´migre´ crime groups is characteristic of organized crime as this condition has developed and come to be identified and defined by American society.     

Efficacy of standard deviational ellipses in the application of criminal geographic profiling

The premise for this study is that the physical and cultural landscape has a deterministic effect on the location and distribution of serial crime. As a consequence, the distribution of linked crime scenes should exhibit a shape and orientation that is consistent with the underlying landscape. Geographic Profiling models that are able to account for these effects will provide more accurate results than those models that do not. Utilizing basic geographic principles of central tendency and spatial diffusion, this research first analyzed the output of circular and elliptical profile models generated for 30 serial burglaries (n = 164) and 67 serial robberies (n = 370) in Baltimore, Maryland between 1994 and 1997. A comparative analysis of the model output reveals that the Standard Deviational Ellipse is significantly (p = 0.000) better able to predict the home location of a serial offender than profiles generated from circles. Next, the relationship between the orientation of elliptical profiles and the mean linear orientation of the corresponding landscape was assessed to reveal a moderate but significant correlation (r = 0.511, p < 0.001). Together, these findings demonstrate that landscape does impact the locations of crime, and is a measurable parameter that can improve the efficacy of geographic profiling. Copyright © 2008 John Wiley & Sons, Ltd.     

The camera perspective bias: a case study

In this case study, we examined how variations of the camera perspective affect adults' assessment of veracity in a real‐life high‐stake situation; a masked male confessing to a very serious crime. The interview, conducted by one of Sweden's most awarded journalists, lasted 30 minutes and consisted of over 100 detailed questions about the crime. The interview was videotaped simultaneously by three cameras positioned at different locations, each taking a unique visual perspective. One camera focused on the suspect only, one on the interviewer only, and one equally on the suspect and the interviewer. Each videotape was shown to 32 adult observers (N = 96) who were asked to assess the suspect's veracity and to report subjective cues justifying their veracity assessment. The results confirmed the camera perspective bias, by showing that the observers in the ‘suspect only’ condition assessed the confession as significantly more reliable than did the observers in the ‘interviewer only’ condition. The observers reported that they relied heavily on verbal cues when assessing the confession, and particularly regarding confidence and consistency. Copyright © 2008 John Wiley & Sons, Ltd.     

Psychological testing for Hispanic Americans

The need for culturally responsive mental health services for Hispanic Americans has never been greater. This population will soon become the largest of all U.S. ethnic minority groups, and recent epidemiologic studies suggest that Hispanics have rates of mental health need that are similar to those found in the general population. Developing culturally relevant mental health services, including diagnostic and assessment procedures, for this population is essential. A synthesis of extant literature pertaining to psychological testing for Hispanics is presented. It is limited to the adult literature and discusses issues related to personality, neuropsychological, and intellectual assessment. The importance of language as it influences the assessment process has also been discussed. Based on clinical and research experiences and the results of the literature review, a set of general guidelines is proposed for those working with this population.